.ipynb .pdf

Binder

Contents

Mice Cortex Analysis

Retrieving and Preparing the Data

#import warnings
#warnings.filterwarnings("ignore")

import numpy as np
import pandas as pd
pd.set_option('display.max_columns', None) 

data = pd.read_csv('Data_Cortex_Nuclear.csv', sep = ',')
data.head()

	MouseID	DYRK1A_N	ITSN1_N	BDNF_N	NR1_N	NR2A_N	pAKT_N	pBRAF_N	pCAMKII_N	pCREB_N	pELK_N	pERK_N	pJNK_N	PKCA_N	pMEK_N	pNR1_N	pNR2A_N	pNR2B_N	pPKCAB_N	pRSK_N	AKT_N	BRAF_N	CAMKII_N	CREB_N	ELK_N	ERK_N	GSK3B_N	JNK_N	MEK_N	TRKA_N	RSK_N	APP_N	Bcatenin_N	SOD1_N	MTOR_N	P38_N	pMTOR_N	DSCR1_N	AMPKA_N	NR2B_N	pNUMB_N	RAPTOR_N	TIAM1_N	pP70S6_N	NUMB_N	P70S6_N	pGSK3B_N	pPKCG_N	CDK5_N	S6_N	ADARB1_N	AcetylH3K9_N	RRP1_N	BAX_N	ARC_N	ERBB4_N	nNOS_N	Tau_N	GFAP_N	GluR3_N	GluR4_N	IL1B_N	P3525_N	pCASP9_N	PSD95_N	SNCA_N	Ubiquitin_N	pGSK3B_Tyr216_N	SHH_N	BAD_N	BCL2_N	pS6_N	pCFOS_N	SYP_N	H3AcK18_N	EGR1_N	H3MeK4_N	CaNA_N	Genotype	Treatment	Behavior	class
0	309_1	0.503644	0.747193	0.430175	2.816329	5.990152	0.218830	0.177565	2.373744	0.232224	1.750936	0.687906	0.306382	0.402698	0.296927	1.022060	0.605673	1.877684	2.308745	0.441599	0.859366	0.416289	0.369608	0.178944	1.866358	3.685247	1.537227	0.264526	0.319677	0.813866	0.165846	0.453910	3.037621	0.369510	0.458539	0.335336	0.825192	0.576916	0.448099	0.586271	0.394721	0.339571	0.482864	0.294170	0.182150	0.842725	0.192608	1.443091	0.294700	0.354605	1.339070	0.170119	0.159102	0.188852	0.106305	0.144989	0.176668	0.125190	0.115291	0.228043	0.142756	0.430957	0.247538	1.603310	2.014875	0.108234	1.044979	0.831557	0.188852	0.122652	NaN	0.106305	0.108336	0.427099	0.114783	0.131790	0.128186	1.675652	Control	Memantine	C/S	c-CS-m
1	309_2	0.514617	0.689064	0.411770	2.789514	5.685038	0.211636	0.172817	2.292150	0.226972	1.596377	0.695006	0.299051	0.385987	0.281319	0.956676	0.587559	1.725774	2.043037	0.445222	0.834659	0.400364	0.356178	0.173680	1.761047	3.485287	1.509249	0.255727	0.304419	0.780504	0.157194	0.430940	2.921882	0.342279	0.423560	0.324835	0.761718	0.545097	0.420876	0.545097	0.368255	0.321959	0.454519	0.276431	0.182086	0.847615	0.194815	1.439460	0.294060	0.354548	1.306323	0.171427	0.158129	0.184570	0.106592	0.150471	0.178309	0.134275	0.118235	0.238073	0.142037	0.457156	0.257632	1.671738	2.004605	0.109749	1.009883	0.849270	0.200404	0.116682	NaN	0.106592	0.104315	0.441581	0.111974	0.135103	0.131119	1.743610	Control	Memantine	C/S	c-CS-m
2	309_3	0.509183	0.730247	0.418309	2.687201	5.622059	0.209011	0.175722	2.283337	0.230247	1.561316	0.677348	0.291276	0.381002	0.281710	1.003635	0.602449	1.731873	2.017984	0.467668	0.814329	0.399847	0.368089	0.173905	1.765544	3.571456	1.501244	0.259614	0.311747	0.785154	0.160895	0.423187	2.944136	0.343696	0.425005	0.324852	0.757031	0.543620	0.404630	0.552994	0.363880	0.313086	0.447197	0.256648	0.184388	0.856166	0.200737	1.524364	0.301881	0.386087	1.279600	0.185456	0.148696	0.190532	0.108303	0.145330	0.176213	0.132560	0.117760	0.244817	0.142445	0.510472	0.255343	1.663550	2.016831	0.108196	0.996848	0.846709	0.193685	0.118508	NaN	0.108303	0.106219	0.435777	0.111883	0.133362	0.127431	1.926427	Control	Memantine	C/S	c-CS-m
3	309_4	0.442107	0.617076	0.358626	2.466947	4.979503	0.222886	0.176463	2.152301	0.207004	1.595086	0.583277	0.296729	0.377087	0.313832	0.875390	0.520293	1.566852	2.132754	0.477671	0.727705	0.385639	0.362970	0.179449	1.286277	2.970137	1.419710	0.259536	0.279218	0.734492	0.162210	0.410615	2.500204	0.344509	0.429211	0.330121	0.746980	0.546763	0.386860	0.547849	0.366771	0.328492	0.442650	0.398534	0.161768	0.760234	0.184169	1.612382	0.296382	0.290680	1.198765	0.159799	0.166112	0.185323	0.103184	0.140656	0.163804	0.123210	0.117439	0.234947	0.145068	0.430996	0.251103	1.484624	1.957233	0.119883	0.990225	0.833277	0.192112	0.132781	NaN	0.103184	0.111262	0.391691	0.130405	0.147444	0.146901	1.700563	Control	Memantine	C/S	c-CS-m
4	309_5	0.434940	0.617430	0.358802	2.365785	4.718679	0.213106	0.173627	2.134014	0.192158	1.504230	0.550960	0.286961	0.363502	0.277964	0.864912	0.507990	1.480059	2.013697	0.483416	0.687794	0.367531	0.355311	0.174836	1.324695	2.896334	1.359876	0.250705	0.273667	0.702699	0.154827	0.398550	2.456560	0.329126	0.408755	0.313415	0.691956	0.536860	0.360816	0.512824	0.351551	0.312206	0.419095	0.393447	0.160200	0.768113	0.185718	1.645807	0.296829	0.309345	1.206995	0.164650	0.160687	0.188221	0.104784	0.141983	0.167710	0.136838	0.116048	0.255528	0.140871	0.481227	0.251773	1.534835	2.009109	0.119524	0.997775	0.878668	0.205604	0.129954	NaN	0.104784	0.110694	0.434154	0.118481	0.140314	0.148380	1.839730	Control	Memantine	C/S	c-CS-m

print ("Rows: ",data.shape[0])
print ("Columns: ",data.shape[1])

Rows:  1080
Columns:  82

An overview of data size and the first 5 rows shows that no information is lost after retrieving from UCI. Column names are:

• 1. Mouse ID

• 2-78. Values of expression levels of 77 proteins; the names of proteins are followed by _N indicating that they were measured in the nuclear fraction.

• 1. Genotype: control (c) or trisomy (t)

• 1. Treatment type: memantine (m) or saline (s)

• 1. Behavior: context-shock (CS) or shock-context (SC)

• 1. Class: c-CS-s, c-CS-m, c-SC-s, c-SC-m, t-CS-s, t-CS-m, t-SC-s, t-SC-m

print("Columns: \n",data.columns.tolist())

Columns: 
 ['MouseID', 'DYRK1A_N', 'ITSN1_N', 'BDNF_N', 'NR1_N', 'NR2A_N', 'pAKT_N', 'pBRAF_N', 'pCAMKII_N', 'pCREB_N', 'pELK_N', 'pERK_N', 'pJNK_N', 'PKCA_N', 'pMEK_N', 'pNR1_N', 'pNR2A_N', 'pNR2B_N', 'pPKCAB_N', 'pRSK_N', 'AKT_N', 'BRAF_N', 'CAMKII_N', 'CREB_N', 'ELK_N', 'ERK_N', 'GSK3B_N', 'JNK_N', 'MEK_N', 'TRKA_N', 'RSK_N', 'APP_N', 'Bcatenin_N', 'SOD1_N', 'MTOR_N', 'P38_N', 'pMTOR_N', 'DSCR1_N', 'AMPKA_N', 'NR2B_N', 'pNUMB_N', 'RAPTOR_N', 'TIAM1_N', 'pP70S6_N', 'NUMB_N', 'P70S6_N', 'pGSK3B_N', 'pPKCG_N', 'CDK5_N', 'S6_N', 'ADARB1_N', 'AcetylH3K9_N', 'RRP1_N', 'BAX_N', 'ARC_N', 'ERBB4_N', 'nNOS_N', 'Tau_N', 'GFAP_N', 'GluR3_N', 'GluR4_N', 'IL1B_N', 'P3525_N', 'pCASP9_N', 'PSD95_N', 'SNCA_N', 'Ubiquitin_N', 'pGSK3B_Tyr216_N', 'SHH_N', 'BAD_N', 'BCL2_N', 'pS6_N', 'pCFOS_N', 'SYP_N', 'H3AcK18_N', 'EGR1_N', 'H3MeK4_N', 'CaNA_N', 'Genotype', 'Treatment', 'Behavior', 'class']

# check data types and missing values
data.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 1080 entries, 0 to 1079
Data columns (total 82 columns):
MouseID            1080 non-null object
DYRK1A_N           1077 non-null float64
ITSN1_N            1077 non-null float64
BDNF_N             1077 non-null float64
NR1_N              1077 non-null float64
NR2A_N             1077 non-null float64
pAKT_N             1077 non-null float64
pBRAF_N            1077 non-null float64
pCAMKII_N          1077 non-null float64
pCREB_N            1077 non-null float64
pELK_N             1077 non-null float64
pERK_N             1077 non-null float64
pJNK_N             1077 non-null float64
PKCA_N             1077 non-null float64
pMEK_N             1077 non-null float64
pNR1_N             1077 non-null float64
pNR2A_N            1077 non-null float64
pNR2B_N            1077 non-null float64
pPKCAB_N           1077 non-null float64
pRSK_N             1077 non-null float64
AKT_N              1077 non-null float64
BRAF_N             1077 non-null float64
CAMKII_N           1077 non-null float64
CREB_N             1077 non-null float64
ELK_N              1062 non-null float64
ERK_N              1077 non-null float64
GSK3B_N            1077 non-null float64
JNK_N              1077 non-null float64
MEK_N              1073 non-null float64
TRKA_N             1077 non-null float64
RSK_N              1077 non-null float64
APP_N              1077 non-null float64
Bcatenin_N         1062 non-null float64
SOD1_N             1077 non-null float64
MTOR_N             1077 non-null float64
P38_N              1077 non-null float64
pMTOR_N            1077 non-null float64
DSCR1_N            1077 non-null float64
AMPKA_N            1077 non-null float64
NR2B_N             1077 non-null float64
pNUMB_N            1077 non-null float64
RAPTOR_N           1077 non-null float64
TIAM1_N            1077 non-null float64
pP70S6_N           1077 non-null float64
NUMB_N             1080 non-null float64
P70S6_N            1080 non-null float64
pGSK3B_N           1080 non-null float64
pPKCG_N            1080 non-null float64
CDK5_N             1080 non-null float64
S6_N               1080 non-null float64
ADARB1_N           1080 non-null float64
AcetylH3K9_N       1080 non-null float64
RRP1_N             1080 non-null float64
BAX_N              1080 non-null float64
ARC_N              1080 non-null float64
ERBB4_N            1080 non-null float64
nNOS_N             1080 non-null float64
Tau_N              1080 non-null float64
GFAP_N             1080 non-null float64
GluR3_N            1080 non-null float64
GluR4_N            1080 non-null float64
IL1B_N             1080 non-null float64
P3525_N            1080 non-null float64
pCASP9_N           1080 non-null float64
PSD95_N            1080 non-null float64
SNCA_N             1080 non-null float64
Ubiquitin_N        1080 non-null float64
pGSK3B_Tyr216_N    1080 non-null float64
SHH_N              1080 non-null float64
BAD_N              867 non-null float64
BCL2_N             795 non-null float64
pS6_N              1080 non-null float64
pCFOS_N            1005 non-null float64
SYP_N              1080 non-null float64
H3AcK18_N          900 non-null float64
EGR1_N             870 non-null float64
H3MeK4_N           810 non-null float64
CaNA_N             1080 non-null float64
Genotype           1080 non-null object
Treatment          1080 non-null object
Behavior           1080 non-null object
class              1080 non-null object
dtypes: float64(77), object(5)
memory usage: 692.0+ KB

data.isnull().sum()

MouseID      0
DYRK1A_N     3
ITSN1_N      3
BDNF_N       3
NR1_N        3
            ..
CaNA_N       0
Genotype     0
Treatment    0
Behavior     0
class        0
Length: 82, dtype: int64

Data types are all reasonable, but missing values exist. There are many protein columns that have 3 missing values, which indicate empty rows? As it only accounts for 0.28% of the rows, it’s safe to remove.

DYRK1A_percent_missing = data['DYRK1A_N'].isnull().sum() * 100 / len(data)
DYRK1A_percent_missing

0.2777777777777778

data1 = data[data['DYRK1A_N'].notna()]
data1.info()

<class 'pandas.core.frame.DataFrame'>
Int64Index: 1077 entries, 0 to 1079
Data columns (total 82 columns):
MouseID            1077 non-null object
DYRK1A_N           1077 non-null float64
ITSN1_N            1077 non-null float64
BDNF_N             1077 non-null float64
NR1_N              1077 non-null float64
NR2A_N             1077 non-null float64
pAKT_N             1077 non-null float64
pBRAF_N            1077 non-null float64
pCAMKII_N          1077 non-null float64
pCREB_N            1077 non-null float64
pELK_N             1077 non-null float64
pERK_N             1077 non-null float64
pJNK_N             1077 non-null float64
PKCA_N             1077 non-null float64
pMEK_N             1077 non-null float64
pNR1_N             1077 non-null float64
pNR2A_N            1077 non-null float64
pNR2B_N            1077 non-null float64
pPKCAB_N           1077 non-null float64
pRSK_N             1077 non-null float64
AKT_N              1077 non-null float64
BRAF_N             1077 non-null float64
CAMKII_N           1077 non-null float64
CREB_N             1077 non-null float64
ELK_N              1062 non-null float64
ERK_N              1077 non-null float64
GSK3B_N            1077 non-null float64
JNK_N              1077 non-null float64
MEK_N              1073 non-null float64
TRKA_N             1077 non-null float64
RSK_N              1077 non-null float64
APP_N              1077 non-null float64
Bcatenin_N         1062 non-null float64
SOD1_N             1077 non-null float64
MTOR_N             1077 non-null float64
P38_N              1077 non-null float64
pMTOR_N            1077 non-null float64
DSCR1_N            1077 non-null float64
AMPKA_N            1077 non-null float64
NR2B_N             1077 non-null float64
pNUMB_N            1077 non-null float64
RAPTOR_N           1077 non-null float64
TIAM1_N            1077 non-null float64
pP70S6_N           1077 non-null float64
NUMB_N             1077 non-null float64
P70S6_N            1077 non-null float64
pGSK3B_N           1077 non-null float64
pPKCG_N            1077 non-null float64
CDK5_N             1077 non-null float64
S6_N               1077 non-null float64
ADARB1_N           1077 non-null float64
AcetylH3K9_N       1077 non-null float64
RRP1_N             1077 non-null float64
BAX_N              1077 non-null float64
ARC_N              1077 non-null float64
ERBB4_N            1077 non-null float64
nNOS_N             1077 non-null float64
Tau_N              1077 non-null float64
GFAP_N             1077 non-null float64
GluR3_N            1077 non-null float64
GluR4_N            1077 non-null float64
IL1B_N             1077 non-null float64
P3525_N            1077 non-null float64
pCASP9_N           1077 non-null float64
PSD95_N            1077 non-null float64
SNCA_N             1077 non-null float64
Ubiquitin_N        1077 non-null float64
pGSK3B_Tyr216_N    1077 non-null float64
SHH_N              1077 non-null float64
BAD_N              864 non-null float64
BCL2_N             792 non-null float64
pS6_N              1077 non-null float64
pCFOS_N            1002 non-null float64
SYP_N              1077 non-null float64
H3AcK18_N          897 non-null float64
EGR1_N             867 non-null float64
H3MeK4_N           807 non-null float64
CaNA_N             1077 non-null float64
Genotype           1077 non-null object
Treatment          1077 non-null object
Behavior           1077 non-null object
class              1077 non-null object
dtypes: float64(77), object(5)
memory usage: 698.4+ KB

#data1[data1['ELK_N', 'MEK_N', 'Bcatenin_N'].notna()]
data2 = data1.dropna(axis=0, subset=('ELK_N', 'MEK_N', 'Bcatenin_N'))
data2.shape

(1047, 82)

data2.info()

<class 'pandas.core.frame.DataFrame'>
Int64Index: 1047 entries, 0 to 1079
Data columns (total 82 columns):
MouseID            1047 non-null object
DYRK1A_N           1047 non-null float64
ITSN1_N            1047 non-null float64
BDNF_N             1047 non-null float64
NR1_N              1047 non-null float64
NR2A_N             1047 non-null float64
pAKT_N             1047 non-null float64
pBRAF_N            1047 non-null float64
pCAMKII_N          1047 non-null float64
pCREB_N            1047 non-null float64
pELK_N             1047 non-null float64
pERK_N             1047 non-null float64
pJNK_N             1047 non-null float64
PKCA_N             1047 non-null float64
pMEK_N             1047 non-null float64
pNR1_N             1047 non-null float64
pNR2A_N            1047 non-null float64
pNR2B_N            1047 non-null float64
pPKCAB_N           1047 non-null float64
pRSK_N             1047 non-null float64
AKT_N              1047 non-null float64
BRAF_N             1047 non-null float64
CAMKII_N           1047 non-null float64
CREB_N             1047 non-null float64
ELK_N              1047 non-null float64
ERK_N              1047 non-null float64
GSK3B_N            1047 non-null float64
JNK_N              1047 non-null float64
MEK_N              1047 non-null float64
TRKA_N             1047 non-null float64
RSK_N              1047 non-null float64
APP_N              1047 non-null float64
Bcatenin_N         1047 non-null float64
SOD1_N             1047 non-null float64
MTOR_N             1047 non-null float64
P38_N              1047 non-null float64
pMTOR_N            1047 non-null float64
DSCR1_N            1047 non-null float64
AMPKA_N            1047 non-null float64
NR2B_N             1047 non-null float64
pNUMB_N            1047 non-null float64
RAPTOR_N           1047 non-null float64
TIAM1_N            1047 non-null float64
pP70S6_N           1047 non-null float64
NUMB_N             1047 non-null float64
P70S6_N            1047 non-null float64
pGSK3B_N           1047 non-null float64
pPKCG_N            1047 non-null float64
CDK5_N             1047 non-null float64
S6_N               1047 non-null float64
ADARB1_N           1047 non-null float64
AcetylH3K9_N       1047 non-null float64
RRP1_N             1047 non-null float64
BAX_N              1047 non-null float64
ARC_N              1047 non-null float64
ERBB4_N            1047 non-null float64
nNOS_N             1047 non-null float64
Tau_N              1047 non-null float64
GFAP_N             1047 non-null float64
GluR3_N            1047 non-null float64
GluR4_N            1047 non-null float64
IL1B_N             1047 non-null float64
P3525_N            1047 non-null float64
pCASP9_N           1047 non-null float64
PSD95_N            1047 non-null float64
SNCA_N             1047 non-null float64
Ubiquitin_N        1047 non-null float64
pGSK3B_Tyr216_N    1047 non-null float64
SHH_N              1047 non-null float64
BAD_N              849 non-null float64
BCL2_N             777 non-null float64
pS6_N              1047 non-null float64
pCFOS_N            972 non-null float64
SYP_N              1047 non-null float64
H3AcK18_N          867 non-null float64
EGR1_N             852 non-null float64
H3MeK4_N           777 non-null float64
CaNA_N             1047 non-null float64
Genotype           1047 non-null object
Treatment          1047 non-null object
Behavior           1047 non-null object
class              1047 non-null object
dtypes: float64(77), object(5)
memory usage: 678.9+ KB

#BAD_N              849 non-null float64
#BCL2_N             777 non-null float64
#pCFOS_N            972 non-null float64
#H3AcK18_N          867 non-null float64
#EGR1_N             852 non-null float64
#H3MeK4_N           777 non-null float64
for col in ['BAD_N', 'BCL2_N', 'pCFOS_N', 'H3AcK18_N', 'EGR1_N', 'H3MeK4_N']:
    percent_missing = data2[col].isnull().sum() * 100 / len(data2)
    print(col, 'percentage of missing values:', percent_missing)

BAD_N percentage of missing values: 18.911174785100286
BCL2_N percentage of missing values: 25.787965616045845
pCFOS_N percentage of missing values: 7.163323782234957
H3AcK18_N percentage of missing values: 17.191977077363898
EGR1_N percentage of missing values: 18.624641833810887
H3MeK4_N percentage of missing values: 25.787965616045845

Number of missing values in columns ‘BAD_N’, ‘BCL2_N’, ‘pCFOS_N’, ‘H3AcK18_N’, ‘EGR1_N’, ‘H3MeK4_N’ are all over 5%. It’s not a good idea to remove NULL in this case. We’ll replace NULL with mean.

for col in ['BAD_N', 'BCL2_N', 'pCFOS_N', 'H3AcK18_N', 'EGR1_N', 'H3MeK4_N']:
    data2[col].fillna(data2[col].mean(), inplace=True)
data2.info()

<class 'pandas.core.frame.DataFrame'>
Int64Index: 1047 entries, 0 to 1079
Data columns (total 82 columns):
MouseID            1047 non-null object
DYRK1A_N           1047 non-null float64
ITSN1_N            1047 non-null float64
BDNF_N             1047 non-null float64
NR1_N              1047 non-null float64
NR2A_N             1047 non-null float64
pAKT_N             1047 non-null float64
pBRAF_N            1047 non-null float64
pCAMKII_N          1047 non-null float64
pCREB_N            1047 non-null float64
pELK_N             1047 non-null float64
pERK_N             1047 non-null float64
pJNK_N             1047 non-null float64
PKCA_N             1047 non-null float64
pMEK_N             1047 non-null float64
pNR1_N             1047 non-null float64
pNR2A_N            1047 non-null float64
pNR2B_N            1047 non-null float64
pPKCAB_N           1047 non-null float64
pRSK_N             1047 non-null float64
AKT_N              1047 non-null float64
BRAF_N             1047 non-null float64
CAMKII_N           1047 non-null float64
CREB_N             1047 non-null float64
ELK_N              1047 non-null float64
ERK_N              1047 non-null float64
GSK3B_N            1047 non-null float64
JNK_N              1047 non-null float64
MEK_N              1047 non-null float64
TRKA_N             1047 non-null float64
RSK_N              1047 non-null float64
APP_N              1047 non-null float64
Bcatenin_N         1047 non-null float64
SOD1_N             1047 non-null float64
MTOR_N             1047 non-null float64
P38_N              1047 non-null float64
pMTOR_N            1047 non-null float64
DSCR1_N            1047 non-null float64
AMPKA_N            1047 non-null float64
NR2B_N             1047 non-null float64
pNUMB_N            1047 non-null float64
RAPTOR_N           1047 non-null float64
TIAM1_N            1047 non-null float64
pP70S6_N           1047 non-null float64
NUMB_N             1047 non-null float64
P70S6_N            1047 non-null float64
pGSK3B_N           1047 non-null float64
pPKCG_N            1047 non-null float64
CDK5_N             1047 non-null float64
S6_N               1047 non-null float64
ADARB1_N           1047 non-null float64
AcetylH3K9_N       1047 non-null float64
RRP1_N             1047 non-null float64
BAX_N              1047 non-null float64
ARC_N              1047 non-null float64
ERBB4_N            1047 non-null float64
nNOS_N             1047 non-null float64
Tau_N              1047 non-null float64
GFAP_N             1047 non-null float64
GluR3_N            1047 non-null float64
GluR4_N            1047 non-null float64
IL1B_N             1047 non-null float64
P3525_N            1047 non-null float64
pCASP9_N           1047 non-null float64
PSD95_N            1047 non-null float64
SNCA_N             1047 non-null float64
Ubiquitin_N        1047 non-null float64
pGSK3B_Tyr216_N    1047 non-null float64
SHH_N              1047 non-null float64
BAD_N              1047 non-null float64
BCL2_N             1047 non-null float64
pS6_N              1047 non-null float64
pCFOS_N            1047 non-null float64
SYP_N              1047 non-null float64
H3AcK18_N          1047 non-null float64
EGR1_N             1047 non-null float64
H3MeK4_N           1047 non-null float64
CaNA_N             1047 non-null float64
Genotype           1047 non-null object
Treatment          1047 non-null object
Behavior           1047 non-null object
class              1047 non-null object
dtypes: float64(77), object(5)
memory usage: 678.9+ KB

C:\Users\Wei\Anaconda3\lib\site-packages\pandas\core\generic.py:6287: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame

See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  self._update_inplace(new_data)

# An overview of the data
data2.describe(include = np.object)

	MouseID	Genotype	Treatment	Behavior	class
count	1047	1047	1047	1047	1047
unique	1047	2	2	2	8
top	3477_6	Control	Memantine	S/C	c-SC-m
freq	1	540	570	537	150

data2.describe(include = np.number).round(3)

	DYRK1A_N	ITSN1_N	BDNF_N	NR1_N	NR2A_N	pAKT_N	pBRAF_N	pCAMKII_N	pCREB_N	pELK_N	pERK_N	pJNK_N	PKCA_N	pMEK_N	pNR1_N	pNR2A_N	pNR2B_N	pPKCAB_N	pRSK_N	AKT_N	BRAF_N	CAMKII_N	CREB_N	ELK_N	ERK_N	GSK3B_N	JNK_N	MEK_N	TRKA_N	RSK_N	APP_N	Bcatenin_N	SOD1_N	MTOR_N	P38_N	pMTOR_N	DSCR1_N	AMPKA_N	NR2B_N	pNUMB_N	RAPTOR_N	TIAM1_N	pP70S6_N	NUMB_N	P70S6_N	pGSK3B_N	pPKCG_N	CDK5_N	S6_N	ADARB1_N	AcetylH3K9_N	RRP1_N	BAX_N	ARC_N	ERBB4_N	nNOS_N	Tau_N	GFAP_N	GluR3_N	GluR4_N	IL1B_N	P3525_N	pCASP9_N	PSD95_N	SNCA_N	Ubiquitin_N	pGSK3B_Tyr216_N	SHH_N	BAD_N	BCL2_N	pS6_N	pCFOS_N	SYP_N	H3AcK18_N	EGR1_N	H3MeK4_N	CaNA_N
count	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000	1047.000
mean	0.428	0.620	0.318	2.295	3.832	0.234	0.183	3.525	0.213	1.434	0.549	0.316	0.316	0.276	0.826	0.729	1.562	1.511	0.445	0.685	0.380	0.363	0.181	1.173	2.471	1.172	0.243	0.273	0.692	0.169	0.405	2.147	0.547	0.452	0.415	0.760	0.585	0.368	0.564	0.357	0.315	0.419	0.389	0.181	0.948	0.161	1.687	0.292	0.429	1.190	0.215	0.167	0.179	0.121	0.156	0.181	0.210	0.121	0.221	0.126	0.527	0.290	1.554	2.238	0.160	1.241	0.847	0.226	0.158	0.135	0.121	0.131	0.445	0.169	0.183	0.206	1.338
std	0.252	0.254	0.050	0.351	0.941	0.041	0.026	1.293	0.033	0.471	0.349	0.048	0.052	0.045	0.119	0.189	0.274	0.478	0.063	0.123	0.219	0.053	0.027	0.337	0.659	0.247	0.031	0.041	0.122	0.028	0.062	0.438	0.283	0.066	0.090	0.124	0.102	0.063	0.089	0.064	0.055	0.068	0.154	0.029	0.172	0.019	0.575	0.038	0.139	0.360	0.187	0.032	0.019	0.014	0.015	0.025	0.069	0.013	0.035	0.027	0.083	0.030	0.249	0.255	0.024	0.174	0.095	0.029	0.027	0.024	0.014	0.023	0.067	0.055	0.037	0.049	0.321
min	0.145	0.245	0.115	1.331	1.738	0.121	0.108	1.344	0.113	0.429	0.149	0.174	0.191	0.146	0.500	0.281	0.302	0.568	0.254	0.335	0.144	0.213	0.114	0.498	1.132	0.151	0.149	0.147	0.199	0.107	0.236	1.135	0.217	0.201	0.228	0.167	0.155	0.226	0.185	0.186	0.195	0.238	0.131	0.118	0.344	0.100	0.599	0.181	0.130	0.529	0.053	-0.062	0.072	0.067	0.100	0.100	0.096	0.086	0.111	0.073	0.284	0.207	0.853	1.322	0.101	0.751	0.577	0.156	0.088	0.081	0.067	0.085	0.259	0.080	0.106	0.102	0.586
25%	0.289	0.475	0.286	2.052	3.143	0.206	0.165	2.448	0.191	1.211	0.339	0.283	0.281	0.245	0.743	0.594	1.374	1.165	0.404	0.600	0.265	0.331	0.162	0.944	1.988	1.021	0.221	0.247	0.612	0.150	0.365	1.822	0.320	0.409	0.351	0.683	0.531	0.326	0.513	0.312	0.275	0.371	0.273	0.160	0.833	0.149	1.272	0.272	0.315	0.925	0.103	0.149	0.168	0.110	0.147	0.167	0.168	0.113	0.195	0.108	0.474	0.269	1.385	2.078	0.143	1.118	0.791	0.206	0.141	0.120	0.110	0.114	0.397	0.134	0.159	0.174	1.076
50%	0.366	0.568	0.315	2.292	3.720	0.232	0.183	3.327	0.211	1.359	0.444	0.323	0.311	0.278	0.821	0.723	1.564	1.355	0.441	0.683	0.327	0.360	0.180	1.096	2.392	1.158	0.245	0.273	0.705	0.167	0.403	2.117	0.446	0.451	0.407	0.762	0.577	0.357	0.562	0.347	0.304	0.407	0.373	0.178	0.935	0.160	1.649	0.294	0.401	1.120	0.147	0.163	0.181	0.122	0.156	0.183	0.188	0.120	0.216	0.123	0.527	0.290	1.528	2.245	0.158	1.237	0.848	0.223	0.158	0.135	0.122	0.129	0.447	0.169	0.183	0.206	1.317
75%	0.489	0.702	0.346	2.529	4.405	0.258	0.198	4.432	0.235	1.565	0.670	0.350	0.350	0.304	0.899	0.853	1.754	1.852	0.483	0.761	0.415	0.394	0.196	1.324	2.871	1.309	0.264	0.301	0.776	0.185	0.442	2.426	0.708	0.488	0.466	0.846	0.635	0.402	0.614	0.395	0.348	0.456	0.474	0.198	1.049	0.172	2.075	0.313	0.540	1.374	0.266	0.178	0.192	0.132	0.165	0.199	0.231	0.128	0.246	0.141	0.580	0.311	1.721	2.427	0.174	1.364	0.917	0.241	0.168	0.140	0.132	0.143	0.491	0.186	0.196	0.219	1.591
max	2.516	2.603	0.497	3.758	8.483	0.539	0.317	7.464	0.306	6.113	3.567	0.493	0.474	0.458	1.408	1.413	2.724	3.061	0.651	1.182	2.133	0.586	0.320	2.803	5.198	2.476	0.387	0.415	1.002	0.305	0.633	3.681	1.873	0.677	0.933	1.125	0.916	0.701	0.972	0.631	0.527	0.722	1.129	0.317	1.680	0.253	3.382	0.817	0.823	2.540	1.459	0.612	0.241	0.159	0.209	0.261	0.603	0.214	0.331	0.537	0.890	0.444	2.586	2.878	0.258	1.897	1.205	0.358	0.282	0.262	0.159	0.257	0.760	0.480	0.361	0.414	2.130

data3 = data2.drop(['MouseID'], axis = 1)
data3.shape

(1047, 81)

# Check Unique values in object feature
object_col = data3.columns[data3.dtypes==object].tolist()

for col in object_col:
    print('The unique values and numbers of', col, 'are:')
    print(data3[col].value_counts())
    print('=========')

The unique values and numbers of Genotype are:
Control    540
Ts65Dn     507
Name: Genotype, dtype: int64
=========
The unique values and numbers of Treatment are:
Memantine    570
Saline       477
Name: Treatment, dtype: int64
=========
The unique values and numbers of Behavior are:
S/C    537
C/S    510
Name: Behavior, dtype: int64
=========
The unique values and numbers of class are:
c-SC-m    150
c-CS-m    150
t-CS-m    135
t-SC-m    135
t-SC-s    132
c-CS-s    120
c-SC-s    120
t-CS-s    105
Name: class, dtype: int64
=========

According to the output and the dataset description, column class contain the information in Genotype, Treatment and Behavior already. Therefore, they can be dropped in data4.

Use data4 for modelling, but we’ll use data3 for the exploration.

data4 = data3.drop(['Genotype', 'Treatment', 'Behavior'], axis = 1)
data4.shape

(1047, 78)

data4.head()

	DYRK1A_N	ITSN1_N	BDNF_N	NR1_N	NR2A_N	pAKT_N	pBRAF_N	pCAMKII_N	pCREB_N	pELK_N	pERK_N	pJNK_N	PKCA_N	pMEK_N	pNR1_N	pNR2A_N	pNR2B_N	pPKCAB_N	pRSK_N	AKT_N	BRAF_N	CAMKII_N	CREB_N	ELK_N	ERK_N	GSK3B_N	JNK_N	MEK_N	TRKA_N	RSK_N	APP_N	Bcatenin_N	SOD1_N	MTOR_N	P38_N	pMTOR_N	DSCR1_N	AMPKA_N	NR2B_N	pNUMB_N	RAPTOR_N	TIAM1_N	pP70S6_N	NUMB_N	P70S6_N	pGSK3B_N	pPKCG_N	CDK5_N	S6_N	ADARB1_N	AcetylH3K9_N	RRP1_N	BAX_N	ARC_N	ERBB4_N	nNOS_N	Tau_N	GFAP_N	GluR3_N	GluR4_N	IL1B_N	P3525_N	pCASP9_N	PSD95_N	SNCA_N	Ubiquitin_N	pGSK3B_Tyr216_N	SHH_N	BAD_N	BCL2_N	pS6_N	pCFOS_N	SYP_N	H3AcK18_N	EGR1_N	H3MeK4_N	CaNA_N	class
0	0.503644	0.747193	0.430175	2.816329	5.990152	0.218830	0.177565	2.373744	0.232224	1.750936	0.687906	0.306382	0.402698	0.296927	1.022060	0.605673	1.877684	2.308745	0.441599	0.859366	0.416289	0.369608	0.178944	1.866358	3.685247	1.537227	0.264526	0.319677	0.813866	0.165846	0.453910	3.037621	0.369510	0.458539	0.335336	0.825192	0.576916	0.448099	0.586271	0.394721	0.339571	0.482864	0.294170	0.182150	0.842725	0.192608	1.443091	0.294700	0.354605	1.339070	0.170119	0.159102	0.188852	0.106305	0.144989	0.176668	0.125190	0.115291	0.228043	0.142756	0.430957	0.247538	1.603310	2.014875	0.108234	1.044979	0.831557	0.188852	0.122652	0.13486	0.106305	0.108336	0.427099	0.114783	0.131790	0.128186	1.675652	c-CS-m
1	0.514617	0.689064	0.411770	2.789514	5.685038	0.211636	0.172817	2.292150	0.226972	1.596377	0.695006	0.299051	0.385987	0.281319	0.956676	0.587559	1.725774	2.043037	0.445222	0.834659	0.400364	0.356178	0.173680	1.761047	3.485287	1.509249	0.255727	0.304419	0.780504	0.157194	0.430940	2.921882	0.342279	0.423560	0.324835	0.761718	0.545097	0.420876	0.545097	0.368255	0.321959	0.454519	0.276431	0.182086	0.847615	0.194815	1.439460	0.294060	0.354548	1.306323	0.171427	0.158129	0.184570	0.106592	0.150471	0.178309	0.134275	0.118235	0.238073	0.142037	0.457156	0.257632	1.671738	2.004605	0.109749	1.009883	0.849270	0.200404	0.116682	0.13486	0.106592	0.104315	0.441581	0.111974	0.135103	0.131119	1.743610	c-CS-m
2	0.509183	0.730247	0.418309	2.687201	5.622059	0.209011	0.175722	2.283337	0.230247	1.561316	0.677348	0.291276	0.381002	0.281710	1.003635	0.602449	1.731873	2.017984	0.467668	0.814329	0.399847	0.368089	0.173905	1.765544	3.571456	1.501244	0.259614	0.311747	0.785154	0.160895	0.423187	2.944136	0.343696	0.425005	0.324852	0.757031	0.543620	0.404630	0.552994	0.363880	0.313086	0.447197	0.256648	0.184388	0.856166	0.200737	1.524364	0.301881	0.386087	1.279600	0.185456	0.148696	0.190532	0.108303	0.145330	0.176213	0.132560	0.117760	0.244817	0.142445	0.510472	0.255343	1.663550	2.016831	0.108196	0.996848	0.846709	0.193685	0.118508	0.13486	0.108303	0.106219	0.435777	0.111883	0.133362	0.127431	1.926427	c-CS-m
3	0.442107	0.617076	0.358626	2.466947	4.979503	0.222886	0.176463	2.152301	0.207004	1.595086	0.583277	0.296729	0.377087	0.313832	0.875390	0.520293	1.566852	2.132754	0.477671	0.727705	0.385639	0.362970	0.179449	1.286277	2.970137	1.419710	0.259536	0.279218	0.734492	0.162210	0.410615	2.500204	0.344509	0.429211	0.330121	0.746980	0.546763	0.386860	0.547849	0.366771	0.328492	0.442650	0.398534	0.161768	0.760234	0.184169	1.612382	0.296382	0.290680	1.198765	0.159799	0.166112	0.185323	0.103184	0.140656	0.163804	0.123210	0.117439	0.234947	0.145068	0.430996	0.251103	1.484624	1.957233	0.119883	0.990225	0.833277	0.192112	0.132781	0.13486	0.103184	0.111262	0.391691	0.130405	0.147444	0.146901	1.700563	c-CS-m
4	0.434940	0.617430	0.358802	2.365785	4.718679	0.213106	0.173627	2.134014	0.192158	1.504230	0.550960	0.286961	0.363502	0.277964	0.864912	0.507990	1.480059	2.013697	0.483416	0.687794	0.367531	0.355311	0.174836	1.324695	2.896334	1.359876	0.250705	0.273667	0.702699	0.154827	0.398550	2.456560	0.329126	0.408755	0.313415	0.691956	0.536860	0.360816	0.512824	0.351551	0.312206	0.419095	0.393447	0.160200	0.768113	0.185718	1.645807	0.296829	0.309345	1.206995	0.164650	0.160687	0.188221	0.104784	0.141983	0.167710	0.136838	0.116048	0.255528	0.140871	0.481227	0.251773	1.534835	2.009109	0.119524	0.997775	0.878668	0.205604	0.129954	0.13486	0.104784	0.110694	0.434154	0.118481	0.140314	0.148380	1.839730	c-CS-m

We end up with a data with 77 descriptive features (proteins) and one target feature (class).

Data exploration

import altair as alt
alt.data_transformers.disable_max_rows()

DataTransformerRegistry.enable('default')

Single column

import matplotlib.pyplot as plt

alt.Chart(data3, width=400).mark_bar().encode(x=alt.X('class', sort='-y'), y='count()').properties(
    title='Number of Measurements for each class')

Target feature values are not quite balanced according to the bar chart.

# Boxplot for column 'DYRK1A_N'
data3.boxplot(column = 'DYRK1A_N')
plt.title("Box Plot Distribution of 'DYRK1A_N' Column")
plt.ylabel('Expression Levels')
plt.show()

The Box Plot Distribution of ‘DYRK1A_N’ Column shows that this protein level distribution is quite skewed. In particular, there are some outliers shown in the box plot. It’s unclear if they are true outliers or just skewed data, so they won’t be removed at this stage. Further investigation is needed to understand this feature more.

for-loop is used to explore the box plot distribution of 8 other columns as below.

for col in ['ITSN1_N', 'BDNF_N', 'NR1_N', 'NR2A_N', 'pAKT_N', 'pERK_N', 'pJNK_N', 'PKCA_N']:
    data3.boxplot(column = col)
    plt.title("Box Plot Distribution of Protein Expression Levels")
    plt.ylabel('Expression Levels')
    plt.show()

alt.Chart(data3).mark_bar().encode(
    alt.X("pBRAF_N", bin=alt.Bin(extent=[0.1, 0.325], step=0.0125)),
    y='count()').properties(title='Histogram distribution of protein expression')

Histogram of pBRAF_N and pCAMKII_N protein expression levels are right-skewed as per the output.

alt.Chart(data3).mark_bar().encode(alt.X('pCAMKII_N', bin=alt.Bin(extent=[1, 7.5], step=0.5)), y = 'count()').properties(
        title='Histogram distribution of protein expression')

Relationship between pairs of attributes

alt.Chart(data3, width=500).mark_boxplot().encode(y='CaNA_N', x='class').properties(
    title='Box Plot of CaNA_N level by Class')

alt.Chart(data3, width=300).mark_boxplot().encode(y='EGR1_N', x='Genotype').properties(
    title='Box Plot of EGR1_N level by Genotype')

alt.Chart(data3, width=300).mark_boxplot().encode(y='SYP_N', x='Behavior').properties(
    title='Box Plot of SYP_N level by Behavior')

alt.Chart(data3, width=300).mark_boxplot().encode(y='H3MeK4_N', x='Treatment').properties(
    title='Box Plot of H3MeK4_N level by Treatment')

alt.Chart(data3).mark_point().encode(x='SNCA_N', y='Ubiquitin_N').properties(
    title='Scatter plot for SNCA_N vs. Ubiquitin_N')

alt.Chart(data3).mark_point().encode(x='PSD95_N', y='pCASP9_N').properties(
    title='Scatter plot for PSD95_N vs. pCASP9_N')

alt.Chart(data3).mark_point().encode(x='P3525_N', y='IL1B_N').properties(
    title='Scatter plot for P3525_N vs. IL1B_N')

alt.Chart(data3).mark_point().encode(x='GluR4_N', y='GluR3_N').properties(
    title='Scatter plot for GluR4_N vs. GluR3_N')

alt.Chart(data3).mark_point().encode(x='GFAP_N', y='Tau_N').properties(
    title='Scatter plot for GFAP_N vs. Tau_N')

alt.Chart(data3).mark_point().encode(x='nNOS_N', y='ERBB4_N').properties(
    title='Scatter plot for nNOS_N vs. ERBB4_N')

Data Modelling using two Classification models

Feature encoding and scaling

# descriptive features
Data = data4.drop(columns = 'class')
# target feature
target = data4['class']

target.shape

(1047,)

target_names = data4['class'].unique()

target.value_counts()

c-SC-m    150
c-CS-m    150
t-CS-m    135
t-SC-m    135
t-SC-s    132
c-CS-s    120
c-SC-s    120
t-CS-s    105
Name: class, dtype: int64

target = target.replace({'c-SC-m': 0, 'c-CS-m':1, 't-SC-m':2, 't-CS-m':3, 't-SC-s':4, 'c-SC-s':5, 'c-CS-s':6, 't-CS-s':7})
target.value_counts()

1    150
0    150
3    135
2    135
4    132
6    120
5    120
7    105
Name: class, dtype: int64

Data.shape

(1047, 77)

from sklearn import preprocessing

Data_df = Data.copy()

Data_scaler = preprocessing.MinMaxScaler()
Data_scaler.fit(Data)
Data = Data_scaler.fit_transform(Data)

pd.DataFrame(Data).head()

	0	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	48	49	50	51	52	53	54	55	56	57	58	59	60	61	62	63	64	65	66	67	68	69	70	71	72	73	74	75	76
0	0.151122	0.212885	0.824638	0.612119	0.630482	0.234015	0.333862	0.168257	0.617322	0.232553	0.157643	0.414113	0.747688	0.482986	0.574775	0.286697	0.650637	0.698164	0.472442	0.619182	0.136915	0.419649	0.317149	0.593715	0.627907	0.596269	0.483960	0.643070	0.766146	0.295596	0.549817	0.747441	0.092035	0.541196	0.152338	0.687269	0.553922	0.467278	0.509996	0.469461	0.436170	0.506018	0.163368	0.323059	0.373254	0.604516	0.303363	0.178458	0.324085	0.402810	0.083584	0.327870	0.690257	0.426816	0.412721	0.477834	0.057168	0.228825	0.531163	0.151103	0.242608	0.169702	0.432843	0.445345	0.044770	0.256699	0.405228	0.162941	0.177312	0.299717	0.426816	0.133930	0.336299	0.087715	0.102890	0.084580	0.705738
1	0.155750	0.188226	0.776455	0.601070	0.585247	0.216807	0.311188	0.154925	0.590173	0.205362	0.159721	0.391151	0.688545	0.432886	0.502766	0.270688	0.587926	0.591606	0.481570	0.590036	0.128911	0.383669	0.291583	0.548032	0.578736	0.584234	0.446943	0.586180	0.724593	0.251840	0.491969	0.701976	0.075589	0.467650	0.137451	0.621033	0.512117	0.409897	0.457694	0.410046	0.383101	0.447496	0.145594	0.322736	0.376914	0.618918	0.302059	0.177452	0.324004	0.386524	0.084514	0.326426	0.664894	0.429952	0.463251	0.488028	0.075103	0.251911	0.576828	0.149555	0.285860	0.212421	0.472327	0.438745	0.054452	0.226088	0.433471	0.220010	0.146494	0.299717	0.429952	0.110434	0.365208	0.080692	0.115874	0.093977	0.749771
2	0.153459	0.205696	0.793572	0.558911	0.575910	0.210527	0.325060	0.153485	0.607102	0.199194	0.154554	0.366797	0.670905	0.434143	0.554483	0.283848	0.590443	0.581558	0.538130	0.566053	0.128651	0.415579	0.292676	0.549983	0.599925	0.580790	0.463293	0.613502	0.730384	0.270560	0.472444	0.710718	0.076445	0.470688	0.137475	0.616142	0.510175	0.375653	0.467725	0.400225	0.356363	0.432379	0.125773	0.334325	0.383316	0.657565	0.332565	0.189744	0.369553	0.373234	0.094486	0.312439	0.700212	0.448652	0.415863	0.475009	0.071718	0.248192	0.607535	0.150434	0.373879	0.202733	0.467603	0.446602	0.044526	0.214719	0.429387	0.186816	0.155920	0.299717	0.448652	0.121560	0.353621	0.080465	0.109050	0.082162	0.868229
3	0.125169	0.157688	0.637326	0.468152	0.480646	0.243717	0.328596	0.132074	0.486945	0.205135	0.127028	0.383877	0.657048	0.537247	0.413245	0.211238	0.522319	0.627585	0.563335	0.463862	0.121509	0.401866	0.319599	0.342080	0.452058	0.545716	0.462966	0.492220	0.667283	0.277208	0.440780	0.536330	0.076936	0.479533	0.144945	0.605654	0.514305	0.338199	0.461189	0.406715	0.402786	0.422989	0.267936	0.220414	0.311501	0.549444	0.364189	0.181101	0.231762	0.333033	0.076248	0.338264	0.669358	0.392700	0.372772	0.397939	0.053257	0.245676	0.562594	0.156083	0.242672	0.184790	0.364359	0.408299	0.119259	0.208943	0.407971	0.179047	0.229602	0.299717	0.392700	0.151031	0.265619	0.126763	0.164241	0.144543	0.721879
4	0.122146	0.157838	0.637787	0.426467	0.441977	0.220323	0.315055	0.129086	0.410194	0.189152	0.117572	0.353282	0.608970	0.422119	0.401706	0.200364	0.486489	0.579839	0.577813	0.416779	0.112407	0.381348	0.297196	0.358746	0.433909	0.519977	0.425815	0.471524	0.627685	0.239874	0.410395	0.519186	0.067645	0.436522	0.121261	0.548237	0.501294	0.283304	0.416698	0.372548	0.353712	0.374357	0.262839	0.212521	0.317400	0.559551	0.376199	0.181805	0.258720	0.337126	0.079697	0.330219	0.686524	0.410187	0.385008	0.422195	0.080162	0.234762	0.656300	0.147044	0.325597	0.187625	0.393332	0.441639	0.116965	0.215528	0.480342	0.245702	0.215008	0.299717	0.410187	0.147711	0.350381	0.096959	0.136298	0.149281	0.812053

df = pd.DataFrame(Data, columns=Data_df.columns)
df.shape

(1047, 77)

df.head()

	DYRK1A_N	ITSN1_N	BDNF_N	NR1_N	NR2A_N	pAKT_N	pBRAF_N	pCAMKII_N	pCREB_N	pELK_N	pERK_N	pJNK_N	PKCA_N	pMEK_N	pNR1_N	pNR2A_N	pNR2B_N	pPKCAB_N	pRSK_N	AKT_N	BRAF_N	CAMKII_N	CREB_N	ELK_N	ERK_N	GSK3B_N	JNK_N	MEK_N	TRKA_N	RSK_N	APP_N	Bcatenin_N	SOD1_N	MTOR_N	P38_N	pMTOR_N	DSCR1_N	AMPKA_N	NR2B_N	pNUMB_N	RAPTOR_N	TIAM1_N	pP70S6_N	NUMB_N	P70S6_N	pGSK3B_N	pPKCG_N	CDK5_N	S6_N	ADARB1_N	AcetylH3K9_N	RRP1_N	BAX_N	ARC_N	ERBB4_N	nNOS_N	Tau_N	GFAP_N	GluR3_N	GluR4_N	IL1B_N	P3525_N	pCASP9_N	PSD95_N	SNCA_N	Ubiquitin_N	pGSK3B_Tyr216_N	SHH_N	BAD_N	BCL2_N	pS6_N	pCFOS_N	SYP_N	H3AcK18_N	EGR1_N	H3MeK4_N	CaNA_N
0	0.151122	0.212885	0.824638	0.612119	0.630482	0.234015	0.333862	0.168257	0.617322	0.232553	0.157643	0.414113	0.747688	0.482986	0.574775	0.286697	0.650637	0.698164	0.472442	0.619182	0.136915	0.419649	0.317149	0.593715	0.627907	0.596269	0.483960	0.643070	0.766146	0.295596	0.549817	0.747441	0.092035	0.541196	0.152338	0.687269	0.553922	0.467278	0.509996	0.469461	0.436170	0.506018	0.163368	0.323059	0.373254	0.604516	0.303363	0.178458	0.324085	0.402810	0.083584	0.327870	0.690257	0.426816	0.412721	0.477834	0.057168	0.228825	0.531163	0.151103	0.242608	0.169702	0.432843	0.445345	0.044770	0.256699	0.405228	0.162941	0.177312	0.299717	0.426816	0.133930	0.336299	0.087715	0.102890	0.084580	0.705738
1	0.155750	0.188226	0.776455	0.601070	0.585247	0.216807	0.311188	0.154925	0.590173	0.205362	0.159721	0.391151	0.688545	0.432886	0.502766	0.270688	0.587926	0.591606	0.481570	0.590036	0.128911	0.383669	0.291583	0.548032	0.578736	0.584234	0.446943	0.586180	0.724593	0.251840	0.491969	0.701976	0.075589	0.467650	0.137451	0.621033	0.512117	0.409897	0.457694	0.410046	0.383101	0.447496	0.145594	0.322736	0.376914	0.618918	0.302059	0.177452	0.324004	0.386524	0.084514	0.326426	0.664894	0.429952	0.463251	0.488028	0.075103	0.251911	0.576828	0.149555	0.285860	0.212421	0.472327	0.438745	0.054452	0.226088	0.433471	0.220010	0.146494	0.299717	0.429952	0.110434	0.365208	0.080692	0.115874	0.093977	0.749771
2	0.153459	0.205696	0.793572	0.558911	0.575910	0.210527	0.325060	0.153485	0.607102	0.199194	0.154554	0.366797	0.670905	0.434143	0.554483	0.283848	0.590443	0.581558	0.538130	0.566053	0.128651	0.415579	0.292676	0.549983	0.599925	0.580790	0.463293	0.613502	0.730384	0.270560	0.472444	0.710718	0.076445	0.470688	0.137475	0.616142	0.510175	0.375653	0.467725	0.400225	0.356363	0.432379	0.125773	0.334325	0.383316	0.657565	0.332565	0.189744	0.369553	0.373234	0.094486	0.312439	0.700212	0.448652	0.415863	0.475009	0.071718	0.248192	0.607535	0.150434	0.373879	0.202733	0.467603	0.446602	0.044526	0.214719	0.429387	0.186816	0.155920	0.299717	0.448652	0.121560	0.353621	0.080465	0.109050	0.082162	0.868229
3	0.125169	0.157688	0.637326	0.468152	0.480646	0.243717	0.328596	0.132074	0.486945	0.205135	0.127028	0.383877	0.657048	0.537247	0.413245	0.211238	0.522319	0.627585	0.563335	0.463862	0.121509	0.401866	0.319599	0.342080	0.452058	0.545716	0.462966	0.492220	0.667283	0.277208	0.440780	0.536330	0.076936	0.479533	0.144945	0.605654	0.514305	0.338199	0.461189	0.406715	0.402786	0.422989	0.267936	0.220414	0.311501	0.549444	0.364189	0.181101	0.231762	0.333033	0.076248	0.338264	0.669358	0.392700	0.372772	0.397939	0.053257	0.245676	0.562594	0.156083	0.242672	0.184790	0.364359	0.408299	0.119259	0.208943	0.407971	0.179047	0.229602	0.299717	0.392700	0.151031	0.265619	0.126763	0.164241	0.144543	0.721879
4	0.122146	0.157838	0.637787	0.426467	0.441977	0.220323	0.315055	0.129086	0.410194	0.189152	0.117572	0.353282	0.608970	0.422119	0.401706	0.200364	0.486489	0.579839	0.577813	0.416779	0.112407	0.381348	0.297196	0.358746	0.433909	0.519977	0.425815	0.471524	0.627685	0.239874	0.410395	0.519186	0.067645	0.436522	0.121261	0.548237	0.501294	0.283304	0.416698	0.372548	0.353712	0.374357	0.262839	0.212521	0.317400	0.559551	0.376199	0.181805	0.258720	0.337126	0.079697	0.330219	0.686524	0.410187	0.385008	0.422195	0.080162	0.234762	0.656300	0.147044	0.325597	0.187625	0.393332	0.441639	0.116965	0.215528	0.480342	0.245702	0.215008	0.299717	0.410187	0.147711	0.350381	0.096959	0.136298	0.149281	0.812053

Feature selection

from sklearn.model_selection import train_test_split
from sklearn.neighbors import KNeighborsClassifier

from sklearn.utils import shuffle
new_Ind = [] 
cur_MaxScore = 0
col_num = 77 
col_Ind_Random = shuffle(range(0, col_num), random_state=1)

for cur_f in range(0, col_num):
    new_Ind.append(col_Ind_Random[cur_f])
    newData = Data[:, new_Ind]
    D_train, D_test, t_train, t_test = train_test_split(newData,
                                                       target,
                                                       test_size = 0.3,
                                                       random_state=0)
    clf = KNeighborsClassifier(5, weights='distance', p=1)
    fit=clf.fit(D_train, t_train)
    cur_Score = clf.score(D_test, t_test)
    if cur_Score < cur_MaxScore:
        new_Ind.remove(col_Ind_Random[cur_f])
    else:
        cur_MaxScore = cur_Score
        print("Score with " + str(len(new_Ind))+" selected features: "+str(cur_Score))

Score with 1 selected features: 0.15555555555555556
Score with 2 selected features: 0.23809523809523808
Score with 3 selected features: 0.29523809523809524
Score with 4 selected features: 0.43174603174603177
Score with 5 selected features: 0.5555555555555556
Score with 6 selected features: 0.6031746031746031
Score with 7 selected features: 0.6317460317460317
Score with 8 selected features: 0.6761904761904762
Score with 9 selected features: 0.7111111111111111
Score with 10 selected features: 0.7428571428571429
Score with 11 selected features: 0.7841269841269841
Score with 12 selected features: 0.8825396825396825
Score with 13 selected features: 0.8857142857142857
Score with 14 selected features: 0.9047619047619048
Score with 15 selected features: 0.9238095238095239
Score with 16 selected features: 0.9428571428571428
Score with 17 selected features: 0.9492063492063492
Score with 18 selected features: 0.9492063492063492
Score with 19 selected features: 0.9555555555555556
Score with 20 selected features: 0.9619047619047619
Score with 21 selected features: 0.9619047619047619
Score with 22 selected features: 0.9650793650793651
Score with 23 selected features: 0.9650793650793651
Score with 24 selected features: 0.9682539682539683
Score with 25 selected features: 0.9746031746031746
Score with 26 selected features: 0.9746031746031746
Score with 27 selected features: 0.9746031746031746
Score with 28 selected features: 0.9777777777777777
Score with 29 selected features: 0.9809523809523809
Score with 30 selected features: 0.9809523809523809
Score with 31 selected features: 0.9809523809523809
Score with 32 selected features: 0.9809523809523809
Score with 33 selected features: 0.9809523809523809
Score with 34 selected features: 0.9873015873015873
Score with 35 selected features: 0.9873015873015873
Score with 36 selected features: 0.9873015873015873
Score with 37 selected features: 0.9873015873015873
Score with 38 selected features: 0.9873015873015873
Score with 39 selected features: 0.9904761904761905
Score with 40 selected features: 0.9904761904761905
Score with 41 selected features: 0.9904761904761905
Score with 42 selected features: 0.9904761904761905
Score with 43 selected features: 0.9904761904761905
Score with 44 selected features: 0.9904761904761905
Score with 45 selected features: 0.9904761904761905
Score with 46 selected features: 0.9904761904761905
Score with 47 selected features: 0.9968253968253968

print("There are " + str(len(new_Ind)) + " features selected:")

There are 47 features selected:

print(new_Ind)

[31, 43, 26, 74, 58, 59, 61, 51, 36, 57, 10, 53, 71, 46, 15, 19, 76, 63, 75, 62, 38, 67, 44, 56, 39, 47, 48, 7, 73, 52, 13, 32, 30, 49, 54, 23, 4, 14, 65, 20, 50, 25, 6, 1, 64, 9, 12]

dataset = pd.DataFrame(Data[:, new_Ind], columns=Data_df.columns[new_Ind])

dataset.shape

(1047, 47)

dataset.head()

	Bcatenin_N	NUMB_N	JNK_N	EGR1_N	GluR3_N	GluR4_N	P3525_N	RRP1_N	DSCR1_N	GFAP_N	pERK_N	ARC_N	pCFOS_N	pPKCG_N	pNR2A_N	AKT_N	CaNA_N	PSD95_N	H3MeK4_N	pCASP9_N	NR2B_N	SHH_N	P70S6_N	Tau_N	pNUMB_N	CDK5_N	S6_N	pCAMKII_N	H3AcK18_N	BAX_N	pMEK_N	SOD1_N	APP_N	ADARB1_N	ERBB4_N	ELK_N	NR2A_N	pNR1_N	Ubiquitin_N	BRAF_N	AcetylH3K9_N	GSK3B_N	pBRAF_N	ITSN1_N	SNCA_N	pELK_N	PKCA_N
0	0.747441	0.323059	0.483960	0.102890	0.531163	0.151103	0.169702	0.327870	0.553922	0.228825	0.157643	0.426816	0.133930	0.303363	0.286697	0.619182	0.705738	0.445345	0.084580	0.432843	0.509996	0.162941	0.373254	0.057168	0.469461	0.178458	0.324085	0.168257	0.087715	0.690257	0.482986	0.092035	0.549817	0.402810	0.412721	0.593715	0.630482	0.574775	0.256699	0.136915	0.083584	0.596269	0.333862	0.212885	0.044770	0.232553	0.747688
1	0.701976	0.322736	0.446943	0.115874	0.576828	0.149555	0.212421	0.326426	0.512117	0.251911	0.159721	0.429952	0.110434	0.302059	0.270688	0.590036	0.749771	0.438745	0.093977	0.472327	0.457694	0.220010	0.376914	0.075103	0.410046	0.177452	0.324004	0.154925	0.080692	0.664894	0.432886	0.075589	0.491969	0.386524	0.463251	0.548032	0.585247	0.502766	0.226088	0.128911	0.084514	0.584234	0.311188	0.188226	0.054452	0.205362	0.688545
2	0.710718	0.334325	0.463293	0.109050	0.607535	0.150434	0.202733	0.312439	0.510175	0.248192	0.154554	0.448652	0.121560	0.332565	0.283848	0.566053	0.868229	0.446602	0.082162	0.467603	0.467725	0.186816	0.383316	0.071718	0.400225	0.189744	0.369553	0.153485	0.080465	0.700212	0.434143	0.076445	0.472444	0.373234	0.415863	0.549983	0.575910	0.554483	0.214719	0.128651	0.094486	0.580790	0.325060	0.205696	0.044526	0.199194	0.670905
3	0.536330	0.220414	0.462966	0.164241	0.562594	0.156083	0.184790	0.338264	0.514305	0.245676	0.127028	0.392700	0.151031	0.364189	0.211238	0.463862	0.721879	0.408299	0.144543	0.364359	0.461189	0.179047	0.311501	0.053257	0.406715	0.181101	0.231762	0.132074	0.126763	0.669358	0.537247	0.076936	0.440780	0.333033	0.372772	0.342080	0.480646	0.413245	0.208943	0.121509	0.076248	0.545716	0.328596	0.157688	0.119259	0.205135	0.657048
4	0.519186	0.212521	0.425815	0.136298	0.656300	0.147044	0.187625	0.330219	0.501294	0.234762	0.117572	0.410187	0.147711	0.376199	0.200364	0.416779	0.812053	0.441639	0.149281	0.393332	0.416698	0.245702	0.317400	0.080162	0.372548	0.181805	0.258720	0.129086	0.096959	0.686524	0.422119	0.067645	0.410395	0.337126	0.385008	0.358746	0.441977	0.401706	0.215528	0.112407	0.079697	0.519977	0.315055	0.157838	0.116965	0.189152	0.608970

D_train, D_test, t_train, t_test = train_test_split(dataset,
                                                   target,
                                                   test_size=0.3,
                                                   stratify=target.values,
                                                   random_state=999)

print(D_train.shape)
print(D_test.shape)
print(t_train.shape)
print(t_test.shape)

(732, 47)
(315, 47)
(732,)
(315,)

KNN

There are several hyperparameters for KNN, for instance

• number of neighbors

• weights

• the type of distance p

Build the KNN classifier with 5 neighbors

KNN_5 = KNeighborsClassifier(5)
fit = KNN_5.fit(D_train, t_train)
t_pre = fit.predict(D_test)

t_pre.shape

(315,)

from sklearn.metrics import confusion_matrix
cm = confusion_matrix(t_test, t_pre)
print(cm)

[[43  0  2  0  0  0  0  0]
 [ 0 45  0  0  0  0  0  0]
 [ 0  0 41  0  0  0  0  0]
 [ 0  0  0 40  0  0  0  1]
 [ 0  0  0  0 40  0  0  0]
 [ 0  0  0  0  0 36  0  0]
 [ 0  1  0  0  0  0 34  1]
 [ 0  0  0  1  0  0  0 30]]

from sklearn.metrics import classification_report
print(classification_report(t_test, t_pre))

              precision    recall  f1-score   support

           0       1.00      0.96      0.98        45
           1       0.98      1.00      0.99        45
           2       0.95      1.00      0.98        41
           3       0.98      0.98      0.98        41
           4       1.00      1.00      1.00        40
           5       1.00      1.00      1.00        36
           6       1.00      0.94      0.97        36
           7       0.94      0.97      0.95        31

    accuracy                           0.98       315
   macro avg       0.98      0.98      0.98       315
weighted avg       0.98      0.98      0.98       315

Hyperparameter tunning

While keeping ‘n_neighbors = 5’, different parameters are set for ‘weights’ and ‘p’.

KNN_5_w = KNeighborsClassifier(5, weights = 'distance')
fit = KNN_5_w.fit(D_train, t_train)
t_pre = fit.predict(D_test)
cm = confusion_matrix(t_test, t_pre)
print(cm)
print(classification_report(t_test, t_pre))

[[45  0  0  0  0  0  0  0]
 [ 0 45  0  0  0  0  0  0]
 [ 0  0 41  0  0  0  0  0]
 [ 0  0  0 40  0  0  0  1]
 [ 0  0  0  0 40  0  0  0]
 [ 0  0  0  0  0 36  0  0]
 [ 0  0  0  0  0  0 35  1]
 [ 0  0  0  1  0  0  0 30]]
              precision    recall  f1-score   support

           0       1.00      1.00      1.00        45
           1       1.00      1.00      1.00        45
           2       1.00      1.00      1.00        41
           3       0.98      0.98      0.98        41
           4       1.00      1.00      1.00        40
           5       1.00      1.00      1.00        36
           6       1.00      0.97      0.99        36
           7       0.94      0.97      0.95        31

    accuracy                           0.99       315
   macro avg       0.99      0.99      0.99       315
weighted avg       0.99      0.99      0.99       315

It seems that weights helped with the accuracy. While keeping’n_neighbors = 5’ and weights, we set p value to 1.

KNN_5_w_1 = KNeighborsClassifier(5, weights = 'distance', p = 1)
fit = KNN_5_w_1.fit(D_train, t_train)
t_pre = fit.predict(D_test)
cm = confusion_matrix(t_test, t_pre)
print(cm)
print(classification_report(t_test, t_pre))

[[45  0  0  0  0  0  0  0]
 [ 0 45  0  0  0  0  0  0]
 [ 0  0 41  0  0  0  0  0]
 [ 0  0  0 41  0  0  0  0]
 [ 0  0  0  0 40  0  0  0]
 [ 1  0  0  0  0 35  0  0]
 [ 0  0  0  0  0  0 35  1]
 [ 0  0  0  1  0  0  0 30]]
              precision    recall  f1-score   support

           0       0.98      1.00      0.99        45
           1       1.00      1.00      1.00        45
           2       1.00      1.00      1.00        41
           3       0.98      1.00      0.99        41
           4       1.00      1.00      1.00        40
           5       1.00      0.97      0.99        36
           6       1.00      0.97      0.99        36
           7       0.97      0.97      0.97        31

    accuracy                           0.99       315
   macro avg       0.99      0.99      0.99       315
weighted avg       0.99      0.99      0.99       315

p=1 or p=2 produces similar accuracy rate. Let’s try lower number of neighbors with p = 2.

KNN = KNeighborsClassifier(4, weights = 'distance', p = 2)
fit = KNN.fit(D_train, t_train)
t_pre = fit.predict(D_test)
cm = confusion_matrix(t_test, t_pre)
print(cm)
print(classification_report(t_test, t_pre))

[[45  0  0  0  0  0  0  0]
 [ 0 45  0  0  0  0  0  0]
 [ 0  0 41  0  0  0  0  0]
 [ 0  0  0 40  0  0  0  1]
 [ 0  0  0  0 40  0  0  0]
 [ 0  0  0  0  0 36  0  0]
 [ 0  0  0  0  0  0 35  1]
 [ 0  0  0  1  0  0  0 30]]
              precision    recall  f1-score   support

           0       1.00      1.00      1.00        45
           1       1.00      1.00      1.00        45
           2       1.00      1.00      1.00        41
           3       0.98      0.98      0.98        41
           4       1.00      1.00      1.00        40
           5       1.00      1.00      1.00        36
           6       1.00      0.97      0.99        36
           7       0.94      0.97      0.95        31

    accuracy                           0.99       315
   macro avg       0.99      0.99      0.99       315
weighted avg       0.99      0.99      0.99       315

Let’s use grid search to see what’s the best number of neighbors to usem

Using grid search method

from sklearn.model_selection import StratifiedKFold, GridSearchCV

cv_method = StratifiedKFold(n_splits=3,shuffle=True, random_state=999)

# define the parameter values
KNN_para = {'n_neighbors': [1, 2, 3, 4, 5,6, 7], 'p': [1, 2]}
KNN_gs = GridSearchCV(KNeighborsClassifier(weights = 'distance'), KNN_para, cv=cv_method, scoring = 'accuracy')

KNN_gs.fit(D_train, t_train)

GridSearchCV(cv=StratifiedKFold(n_splits=3, random_state=999, shuffle=True),
             error_score='raise-deprecating',
             estimator=KNeighborsClassifier(algorithm='auto', leaf_size=30,
                                            metric='minkowski',
                                            metric_params=None, n_jobs=None,
                                            n_neighbors=5, p=2,
                                            weights='distance'),
             iid='warn', n_jobs=None,
             param_grid={'n_neighbors': [1, 2, 3, 4, 5, 6, 7], 'p': [1, 2]},
             pre_dispatch='2*n_jobs', refit=True, return_train_score=False,
             scoring='accuracy', verbose=0)

KNN_gs.best_params_

{'n_neighbors': 1, 'p': 2}

KNN_gs.best_estimator_

KNeighborsClassifier(algorithm='auto', leaf_size=30, metric='minkowski',
                     metric_params=None, n_jobs=None, n_neighbors=1, p=2,
                     weights='distance')

KNN_gs.best_score_

0.98224043715847

KNN_results=pd.DataFrame(KNN_gs.cv_results_)[['mean_test_score', 'std_test_score', 'params']]
KNN_results_sorted=KNN_results.sort_values('mean_test_score', ascending=False)
KNN_results_sorted.head()

	mean_test_score	std_test_score	params
1	0.982240	0.003959	{'n_neighbors': 1, 'p': 2}
3	0.982240	0.003959	{'n_neighbors': 2, 'p': 2}
0	0.978142	0.007871	{'n_neighbors': 1, 'p': 1}
2	0.978142	0.007871	{'n_neighbors': 2, 'p': 1}
7	0.965847	0.010961	{'n_neighbors': 4, 'p': 2}

Using the best method from grid search and test it on testing set

clf = KNeighborsClassifier(algorithm='auto', leaf_size=30, metric='minkowski',
                     metric_params=None, n_jobs=None, n_neighbors=2, p=2,
                     weights='uniform')
fit = clf.fit(D_train, t_train)
t_pre = fit.predict(D_test)
cm = confusion_matrix(t_test, t_pre)
print(cm)
print(classification_report(t_test, t_pre))

[[45  0  0  0  0  0  0  0]
 [ 0 45  0  0  0  0  0  0]
 [ 0  0 41  0  0  0  0  0]
 [ 0  0  0 41  0  0  0  0]
 [ 0  0  0  0 40  0  0  0]
 [ 1  0  0  0  0 35  0  0]
 [ 0  0  0  0  0  0 36  0]
 [ 0  0  0  1  0  0  0 30]]
              precision    recall  f1-score   support

           0       0.98      1.00      0.99        45
           1       1.00      1.00      1.00        45
           2       1.00      1.00      1.00        41
           3       0.98      1.00      0.99        41
           4       1.00      1.00      1.00        40
           5       1.00      0.97      0.99        36
           6       1.00      1.00      1.00        36
           7       1.00      0.97      0.98        31

    accuracy                           0.99       315
   macro avg       0.99      0.99      0.99       315
weighted avg       0.99      0.99      0.99       315

Better results than others. However, t test will need to be conducted to know if this model is significantly better than others. Also, when k =1, the model is likely to be overfitting.

DecisionTree

Train and evaluate the model appropriately.

from sklearn.tree import DecisionTreeClassifier
clf = DecisionTreeClassifier(random_state = 999)
fit = clf.fit(D_train, t_train)
t_pre = fit.predict(D_test)
t_pre.shape

(315,)

cm = confusion_matrix(t_test, t_pre)
print(cm)
print(classification_report(t_test, t_pre))
print(clf.tree_.max_depth)

[[37  0  2  0  3  1  2  0]
 [ 0 41  0  1  0  0  1  2]
 [ 2  0 37  0  0  2  0  0]
 [ 1  3  0 33  0  0  2  2]
 [ 0  0  0  0 39  0  1  0]
 [ 0  0  1  0  0 35  0  0]
 [ 0  3  0  9  0  0 22  2]
 [ 0  6  0  4  0  0  0 21]]
              precision    recall  f1-score   support

           0       0.93      0.82      0.87        45
           1       0.77      0.91      0.84        45
           2       0.93      0.90      0.91        41
           3       0.70      0.80      0.75        41
           4       0.93      0.97      0.95        40
           5       0.92      0.97      0.95        36
           6       0.79      0.61      0.69        36
           7       0.78      0.68      0.72        31

    accuracy                           0.84       315
   macro avg       0.84      0.83      0.83       315
weighted avg       0.84      0.84      0.84       315

12

Set entropy as the criterion

clf = DecisionTreeClassifier(criterion = 'entropy',random_state = 999)
fit = clf.fit(D_train, t_train)
t_pre = fit.predict(D_test)
t_pre.shape
cm = confusion_matrix(t_test, t_pre)
print(cm)
print(classification_report(t_test, t_pre))
print(clf.tree_.max_depth)

[[34  0  4  0  3  1  1  2]
 [ 0 39  0  2  0  0  2  2]
 [ 1  0 40  0  0  0  0  0]
 [ 0  5  0 33  0  0  1  2]
 [ 0  1  0  0 39  0  0  0]
 [ 0  0  1  0  0 35  0  0]
 [ 0  5  0  5  0  0 22  4]
 [ 0  5  0  1  0  0  0 25]]
              precision    recall  f1-score   support

           0       0.97      0.76      0.85        45
           1       0.71      0.87      0.78        45
           2       0.89      0.98      0.93        41
           3       0.80      0.80      0.80        41
           4       0.93      0.97      0.95        40
           5       0.97      0.97      0.97        36
           6       0.85      0.61      0.71        36
           7       0.71      0.81      0.76        31

    accuracy                           0.85       315
   macro avg       0.85      0.85      0.84       315
weighted avg       0.86      0.85      0.85       315

9

clf = DecisionTreeClassifier(criterion = 'entropy',min_samples_split = 10, random_state = 999)
fit = clf.fit(D_train, t_train)
t_pre = fit.predict(D_test)
t_pre.shape
cm = confusion_matrix(t_test, t_pre)
print(cm)
print(classification_report(t_test, t_pre))
print(clf.tree_.max_depth)

[[34  0  3  0  3  1  1  3]
 [ 0 35  0  5  0  0  4  1]
 [ 4  0 37  0  0  0  0  0]
 [ 0  5  0 25  0  0  7  4]
 [ 0  0  0  0 40  0  0  0]
 [ 1  0  1  0  0 34  0  0]
 [ 0  5  0  3  2  0 21  5]
 [ 0  6  0  2  0  0  1 22]]
              precision    recall  f1-score   support

           0       0.87      0.76      0.81        45
           1       0.69      0.78      0.73        45
           2       0.90      0.90      0.90        41
           3       0.71      0.61      0.66        41
           4       0.89      1.00      0.94        40
           5       0.97      0.94      0.96        36
           6       0.62      0.58      0.60        36
           7       0.63      0.71      0.67        31

    accuracy                           0.79       315
   macro avg       0.79      0.79      0.78       315
weighted avg       0.79      0.79      0.79       315

8

Using entropy slightly improve the performance. Now set max_depth as 5.

clf = DecisionTreeClassifier(criterion = 'entropy',max_depth = 5,min_samples_split = 10, random_state = 999)
fit = clf.fit(D_train, t_train)
t_pre = fit.predict(D_test)
t_pre.shape
cm = confusion_matrix(t_test, t_pre)
print(cm)
print(classification_report(t_test, t_pre))
print(clf.tree_.max_depth)

[[34  0  2  0  3  2  4  0]
 [ 0 30  0 11  0  0  2  2]
 [ 4  0 33  0  2  2  0  0]
 [ 0  6  0 27  0  0  5  3]
 [ 0  0  0  0 38  0  0  2]
 [ 0  0  1  0  0 35  0  0]
 [ 0  8  0  8  0  0 17  3]
 [ 0  5  0  9  0  0  0 17]]
              precision    recall  f1-score   support

           0       0.89      0.76      0.82        45
           1       0.61      0.67      0.64        45
           2       0.92      0.80      0.86        41
           3       0.49      0.66      0.56        41
           4       0.88      0.95      0.92        40
           5       0.90      0.97      0.93        36
           6       0.61      0.47      0.53        36
           7       0.63      0.55      0.59        31

    accuracy                           0.73       315
   macro avg       0.74      0.73      0.73       315
weighted avg       0.74      0.73      0.73       315

5

clf = DecisionTreeClassifier(max_depth = 5, min_samples_split = 10, min_samples_leaf=10, random_state = 999)
fit = clf.fit(D_train, t_train)
t_pre = fit.predict(D_test)
t_pre.shape
cm = confusion_matrix(t_test, t_pre)
print(cm)
print(classification_report(t_test, t_pre))
print(clf.tree_.max_depth)

[[32  1  2  0  5  2  0  3]
 [ 0 35  0  6  0  0  1  3]
 [ 2  0 33  0  4  2  0  0]
 [ 0  2  0 28  0  0  0 11]
 [ 0  0  0  0 38  0  0  2]
 [ 0  0  3  0  0 33  0  0]
 [ 0  6  0 10  0  0 14  6]
 [ 0  4  0  7  0  0  1 19]]
              precision    recall  f1-score   support

           0       0.94      0.71      0.81        45
           1       0.73      0.78      0.75        45
           2       0.87      0.80      0.84        41
           3       0.55      0.68      0.61        41
           4       0.81      0.95      0.87        40
           5       0.89      0.92      0.90        36
           6       0.88      0.39      0.54        36
           7       0.43      0.61      0.51        31

    accuracy                           0.74       315
   macro avg       0.76      0.73      0.73       315
weighted avg       0.77      0.74      0.74       315

5

params_DT = {'max_depth': [5, 8, 10, 12], 'min_samples_split': [5, 10, 15], 'min_samples_leaf': [5, 10, 15]}

gs_DT = GridSearchCV(DecisionTreeClassifier(criterion = 'entropy'), 
                          param_grid=params_DT, 
                          cv=cv_method,
                          scoring='accuracy') 

gs_DT.fit(D_train, t_train)

C:\Users\Wei\Anaconda3\lib\site-packages\sklearn\model_selection\_search.py:814: DeprecationWarning: The default of the `iid` parameter will change from True to False in version 0.22 and will be removed in 0.24. This will change numeric results when test-set sizes are unequal.
  DeprecationWarning)

GridSearchCV(cv=StratifiedKFold(n_splits=3, random_state=999, shuffle=True),
             error_score='raise-deprecating',
             estimator=DecisionTreeClassifier(class_weight=None,
                                              criterion='entropy',
                                              max_depth=None, max_features=None,
                                              max_leaf_nodes=None,
                                              min_impurity_decrease=0.0,
                                              min_impurity_split=None,
                                              min_samples_leaf=1,
                                              min_samples_split=2,
                                              min_weight_fraction_leaf=0.0,
                                              presort=False, random_state=None,
                                              splitter='best'),
             iid='warn', n_jobs=None,
             param_grid={'max_depth': [5, 8, 10, 12],
                         'min_samples_leaf': [5, 10, 15],
                         'min_samples_split': [5, 10, 15]},
             pre_dispatch='2*n_jobs', refit=True, return_train_score=False,
             scoring='accuracy', verbose=0)

gs_DT.best_params_

{'max_depth': 12, 'min_samples_leaf': 5, 'min_samples_split': 5}

gs_DT.best_estimator_

DecisionTreeClassifier(class_weight=None, criterion='entropy', max_depth=12,
                       max_features=None, max_leaf_nodes=None,
                       min_impurity_decrease=0.0, min_impurity_split=None,
                       min_samples_leaf=5, min_samples_split=5,
                       min_weight_fraction_leaf=0.0, presort=False,
                       random_state=None, splitter='best')

gs_DT.best_score_

0.7718579234972678

DT_results=pd.DataFrame(gs_DT.cv_results_)[['mean_test_score', 'std_test_score', 'params']]
DT_results_sorted=DT_results.sort_values('mean_test_score', ascending=False)
DT_results_sorted.head()

	mean_test_score	std_test_score	params
27	0.771858	0.004279	{'max_depth': 12, 'min_samples_leaf': 5, 'min_...
20	0.762295	0.007632	{'max_depth': 10, 'min_samples_leaf': 5, 'min_...
9	0.762295	0.010088	{'max_depth': 8, 'min_samples_leaf': 5, 'min_s...
18	0.760929	0.008506	{'max_depth': 10, 'min_samples_leaf': 5, 'min_...
19	0.760929	0.005822	{'max_depth': 10, 'min_samples_leaf': 5, 'min_...

clf = DecisionTreeClassifier(class_weight=None, criterion='entropy', max_depth=8,
                       max_features=None, max_leaf_nodes=None,
                       min_impurity_decrease=0.0, min_impurity_split=None,
                       min_samples_leaf=5, min_samples_split=10,
                       min_weight_fraction_leaf=0.0, presort=False,
                       random_state=None, splitter='best')
fit = clf.fit(D_train, t_train)
t_pre = fit.predict(D_test)
t_pre.shape
cm = confusion_matrix(t_test, t_pre)
print(cm)
print(classification_report(t_test, t_pre))
print(clf.tree_.max_depth)

[[33  0  2  0  3  1  3  3]
 [ 0 35  0  5  0  0  3  2]
 [ 3  0 36  0  0  0  2  0]
 [ 0  5  0 26  0  0  6  4]
 [ 0  2  0  0 38  0  0  0]
 [ 1  0  1  0  0 34  0  0]
 [ 0  8  0  4  0  0 23  1]
 [ 0  5  0  3  0  0  1 22]]
              precision    recall  f1-score   support

           0       0.89      0.73      0.80        45
           1       0.64      0.78      0.70        45
           2       0.92      0.88      0.90        41
           3       0.68      0.63      0.66        41
           4       0.93      0.95      0.94        40
           5       0.97      0.94      0.96        36
           6       0.61      0.64      0.62        36
           7       0.69      0.71      0.70        31

    accuracy                           0.78       315
   macro avg       0.79      0.78      0.78       315
weighted avg       0.79      0.78      0.79       315

8

from sklearn import tree
with open('mice_protein.dot', 'w') as f:
    f=tree.export_graphviz(clf, out_file=f, feature_names = dataset.columns,
                          class_names=target_names, filled=True, rounded=True,
                          special_characters=True)

Jobseeker Analysis